Professional Documents
Culture Documents
ABSTRACT: So far, a general model for the explanation of non-linear shunts and edge currents, often showing
ideality factors n > 2, has been missing. Non-linear shunts like scratches and edge currents are the major source of the
recombination current of industrial crystalline silicon solar cells. Moreover, the reverse current of such cells behaves
always linear or even superlinear instead of saturation-type, as expected. We propose that the edge recombination
current and other non-linear shunts can be described as regions that are highly disturbed, even amorphous, crossing
the pn-junction. The recombination within such regions needs model descriptions that go beyond the SRH
approximation, because the density of defects is so high that carriers recombine via more than one defect (so-called
multi-level recombination). Two variants of coupled defect level recombination, both leading to idealiy factors larger
than 2, are realistically modelled by Sentaurus simulations. Under reverse bias, the high local density of states in the
defective material enables hopping conduction, which explains the linear reverse characteristics.
Keywords: Fundamentals, Silicon, Simulation, Shunts.
1 INTRODUCTION anywhere in the cell area that are responsible for current
with n > 2 [6]. We propose that already the elementary
The Shockley-Read-Hall (SRH) recombination recombination process at the extended defects may be
theory, which assumes recombination via isolated point saturable, which leads to the observed high ideality
defect levels, predicts an ideality factor of the factors. For explaining this saturation, models beyond the
recombination current of n = 2 or less (see e.g. McIntosh SRH approximation have to be considered, which
et al. [1]), and saturation-type reverse characteristics. becomes probable if recombination occurs via extended
Industrial crystalline solar cells usually deviate from this defects [7]. Results of the simulation of two special cases
prediction. Their recombination current is much higher of defect pair recombination, both leading to high ideality
than expected for the given lifetime and often shows an factors, will be introduced.
ideality factor larger than 2 at relative low voltages.
Moreover, their reverse characteristic is linear or even
superlinear. Especially the large ideality factors have 2 EXPERIMENTAL RESULTS
attracted attention already very early [2] and are still
under discussion. Previous explanations of a high ideality 2.1 Artificial shunts, dark forward characteristics
factor were based on trap-assisted tunneling or field-
enhanced recombination [3] via isolated point defect
levels. However, realistic simulations have shown that
the electric field in the depletion region is too low to
enable effective single-level trap-assisted tunneling. Only
if two-level recombination implying at least one shallow
level is considered, which becomes probable only for
high defect concentrations, trap-assisted tunneling
becomes effective and indeed may lead to large ideality
factors [4]. Another explanation of high ideality factors
was based on the emitter series resistance between the
contacts and the recombination site [5]. We will show in
the following that this series resistance is not sufficient to
explain large ideality factors at the observed low
voltages.
A key for the understanding of large ideality factors Fig. 1: AFM image of a 25x25 µm sized region of the
is the fact that such recombination currents are flowing scratch with 27 g load. The scratch is about 1 µm deep.
not homogeneously in the cell, but always at local sites,
e.g. in extended defects. For example, it had been Planar, monocrystalline silicon solar cells with the
realized by many authors that, in the absence of other edge of the pn-junction lying well-passivated below an
defects, the major source of currents with n > 2 is the oxide layer ("PERL"-type cells), behave like "ideal" cells,
edge, where the pn-junction crosses the surface. By lock- i.e. they don't show a recombination current with an
in thermography investigations it had been proven that, ideality factor larger than 2. By cleavage [7, 8] or
besides the edges, there are local non-linear shunts
scratching, typical non-linear shunts can be produced in a curve for the 27 g load.
controlled manner. Here we have made 1 mm long
scratches at the front surface of several 1 cm2 sized 2.2 Artificial shunts, reverse characteristics
"ideal" cells with well-defined loads from 6 to 27 g by Fig. 4 shows the room temperature reverse
using a Vickers indentor at room temperature. In Fig. 1, characteristics of the cells of Figs. 2 and 3 in linear scale.
an AFM image is shown of a scratch performed with 27 g Both the virgin and the 6g scratched cell show a
load. It is apparent that, even for this scratch with the negligible reverse current. We see that the reverse
highest load, the material still was deformed plastically. conductivity strongly correlates with the magnitude of the
In Fig. 2 the dark I-V characteristic and in Fig. 3 the bias- recombination current under forward bias: the higher the
dependent ideality factor of an "ideal" (virgin) cell are recombination current, the higher is the reverse
shown, together with characteristics of 3 equivalent cells, conductivity. The characteristics are nearly linear up to 5
which had been diamond-scratched over 1 mm length V.
0.10
using three different loads. The virgin cell shows an
ideality factor below 1.5 over the whole bias range. The
1E-5
1E-6
1E-6
1E-7 1E-7
1E-8
0.22 0.24 0.26 0.28 0.30
1/4 -1/4
1E-9 1/T [K ]
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Fig. 5: T-depencence of the reverse current of the cell
bias [V] scratched with 27 g load, measured at -1 V reverse bias
Fig. 2: Measured forward I-V characteristics of a virgin
"ideal" cell and 3 equivalent cells, which are diamond The temperature dependence of the I-V
scratched with different loads. Dotted: Rs-corrected curve characteristics may point to the conduction mechanism.
for the 27 g load. In Fig. 5 the T-dependence of the reverse current of the
Rs-corrected cell with the 27 g scratch measured at -1V is shown. This
5.5
load 27g dependence is displayed logarithmically over 1/T1/4. It is
5.0 load 9g found that, at least in the range between -50°C and
4.5 load 6g +100°C, this plot produces a straight line. It is well
virgin known that such a dependence is expected for variable
4.0
ideality factor
ideality factor
energies of the shallow trap level Et and various defect e12=1e16 3.5
1E-6
lifetimes τ [7]. virgin 3.0
1E-7
2.5
1E-8
1E-9 2.0
1E-10 1.5
1E-11 1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6
bias [V]
Fig. 8: Simulated forward I-V characteristics and ideality
factors (dashed) of a cell with a defect modelled by the
Deep DAP recombination model. Both levels are
assumed to lie at midgap.
REFERENCES